Place
Climate, weather, grid mix, local rates, and program geography.
From messy bills to decisions.
Our AI ensemble combines forecasting, building science, and project economics to rank the changes most likely to matter.
Measured context.Calibrated models.Practical decisions.
The model begins with operating reality.
Building
Envelope, HVAC, lighting, refrigeration, motors, and controls.
Operations
Open hours, occupancy, production cycles, and seasonality.
Economics
Tariffs, project cost, maintenance, useful life, and funding.
One estimate is fragile. An ensemble is stronger.
Four complementary models vote, calibrate, and preserve uncertainty instead of hiding it behind one number.
Change-point baseline
Learns how heating, cooling, and base load respond to weather.
Gradient-boosted demand
Finds nonlinear interactions among equipment, schedules, and conditions.
Time-structure model
Captures recurring daily, weekly, and seasonal operating patterns.
Physics guardrails
Keeps sparse-data estimates inside credible building limits.
Baseline first. Savings second.
We compare measured use with a weather- and schedule-matched counterfactual.
Weather-normalized electricity useThe ensemble learns the pre-project pattern, then forecasts a matched post-project baseline under the same weather and operating conditions.
Test the levers before spending the money.
Scenario studio
Adjust three operating levers. Energy, cost, carbon, and payback update together.
9.5 MWhannual electricity avoided
$2,715modeled annual savings
1.8 tCO₂emodeled carbon avoided
4.3 yearssimple payback
Recommendations must clear four gates.
Data
Bills and operating history are cleaned before any estimate is trusted.
Fit
Models are judged on unseen periods, bias, and stability.
Feasibility
Comfort, safety, ventilation, and continuity remain boundaries.
Value
Energy, carbon, maintenance, and economics move together.